Over 25% of expenditures or mental illnesses and substance abuse are due to acute hospitalizations, and these disorders account for 29% of all hospital days and 22% of all hospital costs in the US.1 Acute episodes of illness causing hospitalization are expensive, but many are avoidable with better illness self-management.2 Some interventions for people with serious mental illness (SMI) reduce relapse and re-hospitalization, but they are neither cost effective nor routinely delivered because they require substantial training and involve frequent in-person or telephonic contact. Research demonstrates the effectiveness of automated telehealth devices in reducing crisis-based care for people with unstable medical illnesses. In contrast, much of the research on telehealth for mental illness has focused on providing real-time clinical contact using telephonic or video-based approaches. Although they reduce travel time, these approaches require considerable professional time and expense. Little research has focused on automated telehealth for psychiatric instability in adults at high risk for hospitalization. This study will evaluate the effectiveness of an automated telehealth device monitored by a mental health clinician that allows daily monitoring of symptoms and preemptive intervention to treat early warning signs of relapse to reduce unnecessary acute service use. Pre-post pilot studies by the PI (Pratt) and others, 3, 4 provide evidence of its feasibility and potential to reduce acute service use in peopl with SMI, but a more rigorous evaluation is needed to evaluate effectiveness and investigate potential mechanisms of action. The aims of this study are to compare the effectiveness of 6 months of in-home automated telehealth with health home Usual Care in reducing use and estimated costs of acute services (ER visits and hospitalizations) (Aim 1), and in reducing psychiatric symptom severity (Aim 2) at 6 and 12 months. We will also study potential mechanisms of action (Aim 3). We will randomly assign 300 people with SMI and psychiatric instability who receive services at 1 of 2 community mental health centers to test the following primary hypotheses: Telehealth compared to Usual Care will be associated with less use and lower estimated cost of acute services (H1) and greater improvement in psychiatric symptoms (H2). Also, illness self-management will mediate the relationship between telehealth and psychiatric symptoms, and preemptive provider contact will mediate the relationship between telehealth and acute service use (H3). We will also explore whether telehealth is associated with an increase in all mental health outpatient clinical contacts to collect data for a potential later formal cost analysis. This study addresses the public health challenge of reducing the personal and financial burden caused by psychiatric instability and avoidable use of acute services among people with SMI by evaluating an in-home automated telehealth device with the potential to reduce psychiatric symptoms and decrease acute service use. Findings could substantially enhance consumer quality of life, and may also help inform policy and practice in the management of other vulnerable groups who are hard to engage in traditional clinic-based care.